Study On Warning Algorithm Of Vehicle Collision Avoidance System Based On VANET

The collision accident of automobile has always been the main form of road traffic accident,how to reduce the incidence of collision accident technically has always been an important issue in the field of automobile safety.The traditional collision warning system is mainly forward collision warning system(FCW),this system is based on the stimulus-reaction mechanism.With the rapidly development of short-range wireless communication technology,cooperative collision warning system(CCWS)becomes a new direction to study.Compared with FCW,CCWS can theoretically improve the success rate of collision avoidance because of the collision avoidance behaviors are made by both two drivers.This thesis makes a systematic exploratory study on the warning algorithm of vehicle collision avoidance system based on VANET,the algorithm is divided into four parts: state estimation,trajectory prediction,hazard identification and warning prompt,and corresponding research is done respectively.Vehicle motion estimation is the basic work of vehicle collision warning algorithm based on VANET,it is necessary to do this part because on-board sensors always have some errors.In this part,establish a Carsim\Simulink model to simulate the real process.Firstly,the vehicle parameters,driving conditions and driver control modes are designed in Carsim to simulate the real process of automobile movement.Secondly,output the true state data and adding noise by using the method of Carsim/Simulink simulation,noise selection is based on the the maximum error allowed by collaborative collision warning system,according to the equation of state,write extended kalman filtering program with M language,and import it into the S-function.Finally,run the model,compares the filtered value with the truth value and verifies the validity of the method.The research on trajectory prediction is carried out in combination with hazard identification,the concept of dynamic security boundary is put forward,combining with the safe distance logic algorithm to predict the state of the car after 0.8s.The prediction principle is that after using the extended kalman filter to obtain the real estimation of the automobile motion state,continue to use the equation of state to predict and iterate the calculation,the result is the prediction of the vehicle state after 0.8s.Compares it with the real track,analyzes the cause of error,and puts forward an improved direction.Warning prompt of the collision warning system has to adapt to driving characteristics,in VANET,obtain real-time vehicle driving data become possible.By analyzing the driving data of the car,the driving characteristics of the driver can be determined more and more accurately,and the result can be used to adjust the warning prompt timing.The real driving data of the 19 drivers provided by a car-connected company are samples,using factor analysis and k-means clustering method to realize the driver's driving characteristics identification,this is a sample research of vehicle collision warning system based on VANET.This thesis systematically studies the cooperative collision warning system algorithm,it has some theoretical reference significance for the research and development of the vehicle anti-collision warning system based on VANET.